The Delivery Workforce

The delivery workforce of the study set was early in its career and new to agentic AI, and it learned the agents on live engagements. Individual contributors numbered 667 of the 850 respondents (78.47%). The median career was 2.0 years (n = 797 respondents), and no respondent typed more than one year of agentic AI experience.

Satisfaction, trust and care were associated with every study measure. Respondent and team satisfaction were associated with cancellation more strongly than any of the 100 rated indicators, at r = −0.81 each (n = 150 engagements). Each of the 12 trust and care indicators, toward the financial institution, the manager and the employer, was associated with all four measures. Regular contact with the financial institution was the exception, and its associations were equivalent to zero.

Intent to leave was high and concentrated on canceled engagements. A rating of 7 or above on likelihood to leave was given by 430 of 798 respondents (53.88%). Every rating from a respondent on a canceled engagement was distressed, 300 of 300.

“two people on my team left already. one of them was training me. i am four months in and already worn out in a way i did not expect from an office job.”Individual Contributor, BPO, Corporate and back-office functions

The workforce described in these sections performs the manual alterations of the delivery team. Its knowledge of the corrections was held in people and not in documents, and the knowledge leaves with the people who hold it (Part 5).

As in the other parts of the study, the evidence comes from vendor delivery staff, and no financial institution was approached or asked for comment.

The respondents were early in their careers, and every role was new to agentic AI. Individual contributors numbered 667 of the 850 respondents (78.47%), engagement managers 150 (17.65%), portfolio directors 31 (3.65%) and customer management executives 2 (0.24%). Each engagement had one engagement manager among its respondents and a majority of individual contributors.

Career experience was short. The mean career was 3.9 years and the median 2.0 years (n = 797 respondents). Experience in the outsourcing industry had a mean of 2.6 years and a median of 1.0 (n = 806 respondents). Experience in the financial industry had a mean of 2.1 years and a median of 1.0 (n = 789 respondents).

Agentic AI experience was short in every role. No respondent typed more than one year of agentic AI experience, and 467 of the 793 respondents who answered (58.89%) typed zero, recorded as under one year. Career experience followed seniority and agentic AI experience did not. Engagement managers and portfolio directors had longer careers than individual contributors and no more experience of agentic AI.

“I might not have been working as long as them but I have more AI experience than my entire management team combined.”Individual Contributor, BPO, Payments, cards and disputes

The analysis of experience has a firm limit. Experience tracks role, and role was associated with the respondents’ own ratings and with their intent to leave. The small respondent-level associations of experience with intent to leave therefore cannot be separated from role. At engagement level, the engagement means of career, outsourcing and financial-industry experience were each equivalent to zero against cancellation, customer satisfaction and the engagement indicator mean.

1.1 Role Composition of the Engagements

The role structure of the respondents was fixed by the structure of the teams. Every engagement had individual contributors as the majority of its respondents, and no engagement-level test could compare one role against another. Role is therefore examined at respondent level, where it was associated with the respondents’ own ratings and with their intent to leave.

1.1.1 Role Composition Fixed by the Structure of the Teams

All 850 respondents answered the question on their role. Individual contributors made up more than three quarters of the respondents and were present on each of the 150 engagements, as were engagement managers. A portfolio director was present on 31 of the 150 engagements (20.67%).

The structure of the teams fixed the analysis. Individual Contributor was the most frequent answer on every engagement, and no engagement-level test across roles could be computed. At respondent level, with standard errors clustered by engagement, role was associated with the respondents’ own ratings and with their intent to leave (p < 0.001 for each). Intent to leave at 7 or above on the 10-point scale was reported by 342 of 616 individual contributors (55.52%), 75 of 150 engagement managers (50.00%) and 13 of 31 portfolio directors (41.94%). Cancellation and customer satisfaction, both properties of the engagement, did not differ by role.

The written responses on role numbered 22, from 22 respondents on 21 engagements, and 7 were negative. Six of the seven negative responses described a decision that the respondent judged to lie beyond the respondent’s role, made because neither firm would make it.

“I end up being the person who decide who owns what, and that is not really my job level. My team lead take it fine but I notice the newer guys ask me instead of asking the client side, so now im the middle man for surveillance rule questions.”Individual Contributor, ITO, Regulatory, audit, tax and legal

“Same story every week, nobody wants to be the one who says yes so I decide. Its not my job to own that call but somebody has to”Individual Contributor, CXM, Technology engineering and operations

“I dont say much on this one, but it cost me sleep some week, numbers move and I am the one who decide what go to client file”Individual Contributor, BPO, Corporate and back-office functions

1.2 Career, Outsourcing, and Financial Industry Experience

The respondents had short careers in each of the three measures of experience the survey recorded. The medians were 2.0 years of career, 1.0 year in the outsourcing industry and 1.0 year in the financial industry, on bases of 797, 806 and 789 respondents. An answer of zero, recorded as under one year, came from 61 respondents for career, 112 for the outsourcing industry and 136 for the financial industry.

Experience was not associated with the engagement’s outcomes. For each of the three measures, the engagement mean of experience was equivalent to zero against cancellation, customer satisfaction and the engagement indicator mean. At respondent level, more experienced respondents reported slightly lower intent to leave, with correlations between r = −0.06 and r = −0.09, each small enough to be equivalent to zero. Each measure of experience tracks role, and the associations cannot be separated from role.

1.2.1 Career Experience of the Delivery Workforce

Respondents typed the number of years of experience in their career. The mean answer was 3.9 years and the median 2.0, from 0 to 25 (n = 797 respondents). An answer of zero, recorded as under one year, came from 61 respondents. Career experience followed role, with a mean of 1.9 years among individual contributors (n = 614 respondents), 10.0 among engagement managers (n = 150) and 15.1 among portfolio directors (n = 31).

The engagement mean of career experience was not associated with cancellation (r = −0.11, p = 0.191, n = 150 engagements). The association with customer satisfaction among active engagements was also absent (r = 0.03, p = 0.757, n = 94 engagements). The engagement indicator mean showed no association either (r = 0.10, p = 0.208, n = 150 engagements), and each of the three associations was equivalent to zero. At respondent level, more experienced respondents reported lower intent to leave (r = −0.09, cluster bootstrap 95% CI −0.15 to −0.04, p = 0.001, n = 746 respondents). The association was small enough to be equivalent to zero.

Four written responses addressed career experience, a base of anecdote strength, and three came from respondents early in their careers.

“Im a fresher and know more about AI than anyone in our executive team.”Individual Contributor, KPO, Corporate and back-office functions

1.2.2 Outsourcing Industry Experience

Respondents typed the number of years they had worked in the outsourcing industry. The mean answer was 2.6 years and the median 1.0, from 0 to 17 (n = 806 respondents). Under one year, typed as zero, was the answer of 112 respondents.

The engagement mean of outsourcing experience was not associated with cancellation (r = −0.04, p = 0.617, n = 150 engagements). The association with customer satisfaction among active engagements was also absent (r = 0.06, p = 0.573, n = 94 engagements). The engagement indicator mean showed no association either (r = 0.07, p = 0.415, n = 150 engagements), and each of the three associations was equivalent to zero. At respondent level, more experienced respondents reported lower intent to leave (r = −0.07, cluster bootstrap 95% CI −0.14 to −0.01, p = 0.025, n = 758 respondents). The association was small enough to be equivalent to zero and is bound to role.

1.2.3 Financial Industry Experience

Respondents typed the number of years they had worked within the financial industry. The mean answer was 2.1 years and the median 1.0, from 0 to 15 (n = 789 respondents). Zero, recorded as under one year, was typed by 136 respondents.

The engagement mean of financial-industry experience was not associated with cancellation (r = −0.03, p = 0.684, n = 150 engagements). The association with customer satisfaction among active engagements was also absent (r = −0.08, p = 0.470, n = 94 engagements). The engagement indicator mean showed no association either (r = 0.00, p = 0.955, n = 150 engagements), and each of the three associations was equivalent to zero. At respondent level, the association with intent to leave (r = −0.06, cluster bootstrap 95% CI −0.12 to 0.00, p = 0.033, n = 742 respondents) was equivalent to zero. It failed three robustness checks.

A single written response addressed the indicator, an anecdote. It recorded a team without banking experience learning the financial tasks from tickets.

“Send us on proper payroll training first… nobody in my team ever worked in bank before, we learn from tickets”Individual Contributor, BPO, Corporate and back-office functions

1.3 Agentic AI Experience

Agentic AI experience was short in every role, and the current engagement was the first or second agentic AI outsourcing engagement for three quarters of the respondents. Seniority added years of career and no years of agentic AI. The mean typed answer for agentic AI experience was 0.4 years among individual contributors (n = 611 respondents), 0.5 among engagement managers (n = 150) and 0.3 among portfolio directors (n = 30). Each answer was zero, recorded as under one year, or one year.

The count of agentic engagements gives the same result on a second measure. The current engagement was the first agentic engagement for 371 respondents (46.43%) and the second for 238 (29.79%), together 76.22% of the 799 who answered. Neither the years of agentic AI experience nor the count of engagements was associated with any of the four study measures.

1.3.1 Agentic AI Experience Capped at One Year in Every Role

Respondents typed the number of years of experience they had with agentic AI. Zero, recorded as under one year, was typed by 467 of the 793 respondents who answered (58.89%), and one year by 326 (41.11%). No respondent typed more than one year. The answer did not differ by role. Agentic AI experience, unlike career experience, does not track seniority.

The engagement mean of agentic AI experience was not associated with cancellation (r = 0.00, p = 0.993, n = 150 engagements). The association with customer satisfaction among active engagements was also absent (r = 0.13, p = 0.220, n = 94 engagements). The engagement indicator mean showed no association either (r = 0.02, p = 0.829, n = 150 engagements). At respondent level, agentic AI experience was not associated with intent to leave (r = −0.04, p = 0.232, n = 746 respondents).

Each of the four associations was equivalent to zero, and every robustness check held. No segment family recorded in the study separated the respondents on agentic AI experience.

1.3.2 Agentic Engagements Worked Across the Career

Respondents typed the number of agentic AI outsourcing engagements they had worked on across their careers, including the current one. The mean answer was 1.8 and the median 2.0, from 1 to 5 (n = 799 respondents). The current engagement was the first for 371 respondents (46.43%). Engagement managers had worked on a mean of 2.5 such engagements (n = 150 respondents) and portfolio directors 2.4 (n = 31), against 1.7 for individual contributors (n = 617).

The engagement mean of the count was not associated with cancellation (r = −0.05, n = 150 engagements) or with customer satisfaction among active engagements (r = −0.02, n = 94 engagements). The engagement indicator mean showed no association either (r = 0.07, n = 150 engagements). At respondent level it was not associated with intent to leave (r = −0.03, n = 751 respondents). Each association was equivalent to zero, and every robustness check held.

Eight written responses addressed the indicator, a base of anecdote strength. Respondents with several engagements behind them judged the current one against earlier ones, and respondents on their first could not.

“I have seen one engagement only. What I can say is I got no sanctions training at all before I started, I learnt the lists from a colleague.”Individual Contributor, KPO, Financial crime and client onboarding

“i have run four of these now. on the account before this one, a card operations desk, we had a standing call every tuesday with one person from their side who could say yes or no on the spot, and if she said no it was no and we moved on.”Engagement Manager, CXM, Corporate and back-office functions

1.4 Self-Assessed Knowledge and Expertise

The respondents rated their own knowledge of the financial industry, their knowledge of agentic AI in the financial industry and their overall agentic AI expertise. Each had about a quarter of its answers in the distressed range. The share ranged from 217 of 796 answers (27.26%) for the financial industry to 225 of 797 (28.23%) for agentic AI in the financial industry.

Each self-assessment was associated with all four study measures, and every association held under each of the 22 robustness checks. The association with the engagement indicator mean ranged from r = 0.78 to r = 0.80 (n = 150 engagements), although none of the three is a rated indicator. Respondents on canceled engagements rated their own knowledge and expertise far lower than respondents on active engagements. The pattern is consistent with self-assessments that record the respondent’s standing on the engagement as much as a fixed level of knowledge.

The written responses describe knowledge built on the engagement and not brought to it. Respondents arrived without the financial domain, received days of generic training and learned from colleagues on live work. The same knowledge, once gained, was what many of the respondents who wrote valued most in the job.

“None of us came from wealth management, so when the tool flags an advisor call for suitability we are reading the rule for the first time ourselves. Thus people stay late going through the clients policy pack on their own time. Nobody trained us for this, we taught each other”Individual Contributor, CXM, Credit, underwriting and suitability

1.4.1 Self-Rated Knowledge of the Financial Industry

Respondents rated how well they know the financial industry. The mean rating was 4.8 and the median 5.0 (n = 796 respondents), and 217 of the 796 answers (27.26%) were distressed. Respondents on canceled engagements gave a mean of 3.5 (n = 309 respondents), far lower than in the three active stages.

The rating was associated with all four study measures. Higher ratings were associated with less cancellation (r = −0.61, n = 150 engagements) and higher customer satisfaction among active engagements (r = 0.77, n = 94 engagements). They were also associated with a higher engagement indicator mean (r = 0.80, n = 150 engagements) and lower intent to leave (r = −0.76, n = 750 respondents). Every association held under each of the 22 robustness checks.

The 30 written responses came from the comment field on the employer’s capability (16) and the question on what the respondent loves most about the job (14). Positive responses numbered 18 and negative 10. Respondents in the first group arrived without the financial domain and learned it from colleagues on live work, and respondents in the second named that domain knowledge among what they valued.

“I come on this desk knowing nothing about settlement, they give me two e-learning module and a recording, and after that i am on live breaks in same week. My senior teach me, not company.”Individual Contributor, BPO, Trade, settlement and reconciliation

“i had one week of a slide deck about banking and then i am on a private equity reporting job with no idea what a capital call is. laptop and access come quick, that part they did well, i give them that. everything else i learn from a colleague who is also new.”Individual Contributor, KPO, Corporate and back-office functions

1.4.2 Self-Rated Knowledge of Agentic AI in the Financial Industry

Respondents rated how well they know agentic AI specific to the financial industry. The mean rating was 4.9 and the median 5.0 (n = 797 respondents), and 225 of the 797 answers (28.23%) were distressed. Respondents on canceled engagements gave a mean of 3.6 (n = 306 respondents).

All four of its study-measure associations held under each of the 22 robustness checks. Higher ratings were associated with less cancellation (r = −0.57, n = 150 engagements) and higher customer satisfaction among active engagements (r = 0.76, n = 94 engagements). They were also associated with a higher engagement indicator mean (r = 0.78, n = 150 engagements) and lower intent to leave (r = −0.75, n = 748 respondents).

Seven written responses addressed the indicator, a base of anecdote strength, and all seven were positive. Each named agentic AI work in a financial domain learned on the engagement. Three placed that learning beside an engagement that was shut down, terminated or held together with difficulty.

“learned agentic payroll operations … team stayed steady through the shutdown”Engagement Manager, BPO, Corporate and back-office functions

1.4.3 Self-Rated Agentic AI Expertise

Respondents rated their overall agentic AI expertise. The mean rating was 4.9 and the median 5.0 (n = 800 respondents), and 222 of the 800 answers (27.75%) were distressed. On canceled engagements the mean was 3.5 (n = 308 respondents).

Each of its four associations survived all 22 robustness checks. Respondents who rated their expertise higher were on engagements with less cancellation (r = −0.61, n = 150 engagements) and higher customer satisfaction among active engagements (r = 0.74, n = 94 engagements). The rating also tracked a higher engagement indicator mean (r = 0.78, n = 150 engagements) and lower intent to leave (r = −0.78, n = 753 respondents).

The 37 written responses came from the question on what the respondent loves most about the job (30) and the comment field on the employer’s capability (7). Positive responses numbered 24 and negative 4. Every response to the first question named agentic AI skill learned on the engagement, and the skill named most was the quality assurance of agent output. The responses on the employer described training of days or videos, and respondents who knew more about AI than the people senior to them.

“The technical stretch of learning how to evaluate an agent’s own output”Engagement Manager, ITO, Technology engineering and operations

“… was my first job out of university. They sent me to training. I knew more about AI than the trainer.”Individual Contributor, BPO, Corporate and back-office functions

1.5 Certifications and Sources of AI Knowledge

Professional certifications divided along role without overlap, and the answer restates role. AI certifications were held by fewer than half of the respondents who answered, and the six sources of AI knowledge were named in almost equal numbers. None of the three separated the engagements on any study measure.

The written responses on the sources of AI knowledge describe a thin formal route. The 37 responses came from individual contributors on 33 engagements, and 19 were negative. Whatever source they named, the respondents who wrote described learning the agents from colleagues, from outside material and on live work, after training of slides and videos.

“what they did is put five of us on agent test automation with nobody who has built it before, my lead learn it from the vendor documents same as me. we had one trainer for whole floor and his diary is full for weeks.”Individual Contributor, ITO, Technology engineering and operations

1.5.1 Professional Certifications Divided by Role

Respondents named their professional certifications, and 793 of the 850 answered. The answer None was given by 236 respondents (29.76%), followed by Six Sigma Green Belt (127), ITIL Foundation (126) and AWS Certified Cloud Practitioner (124). Each answer was given by respondents of a single role.

Individual contributors gave None or one of the first three certifications. Engagement managers gave PMP alone, PMP with ITIL or Scrum Master, or Six Sigma Black Belt. Portfolio directors gave PMP with an MBA or ITIL Expert, or PgMP.

The certification assigned to each engagement was not associated with cancellation (chi-square 4.29, p = 0.235, n = 99 engagements). It was also not associated with customer satisfaction among active engagements (Welch F 0.00, p = 0.992, n = 50 engagements). The engagement indicator mean (Welch F 1.60, p = 0.211, n = 99 engagements) and intent to leave (Welch F 1.15, p = 0.346, n = 99 engagements) showed no association either. The respondent-level differences follow the order of the roles and cannot be separated from role.

1.5.2 Artificial Intelligence Certifications

Respondents named their AI certifications, and 788 of the 850 answered. The answer None was given by 437 of the 788 respondents (55.46%). The other 351 named one of five AI certifications, each held by between 66 and 78 respondents.

The answers did not differ by role, vendor type or any segment family recorded at engagement level. With None included as a level, the answers did not separate the respondents on cancellation, customer satisfaction, the engagement indicator mean or intent to leave.

1.5.3 Sources of Agentic AI Knowledge

Respondents named where they gained their AI knowledge, and 800 of the 850 answered. The six sources were named in almost equal numbers, from 125 to 140 respondents each, and the source did not differ by role, vendor type or any segment family recorded at engagement level.

The source assigned to each engagement was not associated with cancellation (chi-square 2.39, p = 0.792, n = 94 engagements). The association with customer satisfaction among active engagements was also absent (Welch F 0.43, p = 0.521, n = 22 engagements). The engagement indicator mean (Welch F 0.24, p = 0.940, n = 94 engagements) and intent to leave (Welch F 0.36, p = 0.871, n = 94 engagements) showed no association either. The results for cancellation, the engagement indicator mean and intent to leave were equivalent to zero. The result for customer satisfaction, on 22 engagements, was not shown to be equivalent to zero, and every robustness check held.

The 37 written responses came from individual contributors on 33 engagements, and 19 were negative and 14 positive. The answer to the question records where knowledge was said to come from. The written responses record a formal route of slides and videos, followed by learning from colleagues on live work.

“i did the agent course after i was already three months on the desk, therefore i learned the hard way first, then the course told me what i should have done. Order is wrong, content is fine”Individual Contributor, BPO, Corporate and back-office functions

The immediate manager was the subject named most in the answers on what respondents love most about the job, in 169 of 786 answers. Written responses on the manager’s effectiveness numbered 248, from 226 respondents on 112 engagements, and 213 were positive and 30 negative. Respondents who rated the engagement low placed the gap in the management line senior to their own lead.

“My lead is good, he pushes back where he can. But nobody above him has built one of these either, and thats the real gap.”Individual Contributor, ITO, Regulatory, audit, tax and legal

“A manager who pushed back on the vendor honestly, to the end”Individual Contributor, KPO, Regulatory, audit, tax and legal

The ratings of the manager were associated with every study measure. Each of the five manager indicators was distressed in between 241 and 253 answers, on bases of 791 to 803. Each was associated with cancellation, customer satisfaction, the engagement indicator mean and intent to leave. No association of the five indicators failed a robustness check.

The written evidence separates the manager in front of the team from the knowledge of the management line senior to the manager. Written responses on the manager’s knowledge of agentic AI numbered 10, and all 10 were negative. Respondents described questions that a lead passed upward and that did not return, and a management chain that had not run agents before the engagement.

2.1 Manager Effectiveness and Knowledge

The five manager indicators cover the manager’s effectiveness, knowledge of the financial industry, knowledge of agentic AI, and ability to manage agentic outsourcing engagements, in general and for global financial institutions. Each had a mean rating of 4.7 or 4.8 and a median of 5.0, on bases of 791 to 803 respondents. The distressed share ranged from 241 of 800 answers (30.13%) to 253 of 800 (31.63%).

Correlations with cancellation and the engagement indicator mean are computed on 150 engagements, with customer satisfaction on the 94 active engagements, and with intent to leave on between 745 and 753 respondents.

The five indicators moved together and cannot be ranked against one another. Their correlations with the engagement indicator mean lay between r = 0.82 and r = 0.84 (n = 150 engagements), although none of the five is one of the rated indicators that compose it. Respondents on canceled engagements rated each between 3.2 and 3.3, on bases of 296 to 304 respondents.

2.1.1 Effectiveness of the Immediate Manager

Respondents rated whether their manager is effective. The mean rating was 4.8 and the median 5.0 (n = 802 respondents), and 242 of the 802 answers (30.17%) were distressed. Higher ratings were associated with less cancellation (r = −0.67, n = 150 engagements) and lower intent to leave (r = −0.68, n = 753 respondents). The rating differed by engagement status and lifecycle stage, with a mean of 3.2 on canceled engagements (n = 304 respondents).

The 248 written responses came from 226 respondents on 112 engagements, and 169 answered the question on what the respondent loves most about the job. Positive responses numbered 213 and negative 30. Respondents named their manager among what they valued, including on engagements that failed.

“strong manager support / meaningful compliance work / steady client trust”Engagement Manager, ITO, Regulatory, audit, tax and legal

“My team lead is doing what she can with what she got, she stay late same as us. Nobody trained anybody for this, they just put HR cases in front of us and said use the tool. Thats not her fault.”Individual Contributor, BPO, Corporate and back-office functions

Where respondents criticized a manager, the criticism was absence or an escalation that never happened. Active engagements with a negative written response had lower customer satisfaction, and respondents whose written responses were more negative reported higher intent to leave.

“my manager is on another account most of the time so there’s nobody to ask.”Individual Contributor, CXM, Corporate and back-office functions

2.1.2 Manager Knowledge of the Financial Industry

Respondents rated whether their manager knows the financial industry well. The mean rating was 4.8 and the median 5.0 (n = 800 respondents), and 253 of the 800 answers (31.63%) were distressed, the highest share among the five manager indicators. Higher ratings were associated with less cancellation (r = −0.62, n = 150 engagements) and higher customer satisfaction among active engagements (r = 0.70, n = 94 engagements). The mean on canceled engagements was 3.2 (n = 302 respondents).

The one written response on the indicator is an anecdote. It recorded leads learning the financial domain at the same time as the people they led.

“Where they are behind is insurence itself, most of our leads never touched policy admin before this, they are learning it same time we are. Can’t blame them for that really.”Individual Contributor, ITO, Client servicing and communications

2.1.3 Manager Knowledge of Agentic AI

Respondents rated whether their manager knows agentic AI well. The mean rating was 4.7 and the median 5.0 (n = 803 respondents), and 242 of the 803 answers (30.14%) were distressed. Higher ratings were associated with higher customer satisfaction among active engagements (r = 0.72, n = 94 engagements) and a higher engagement indicator mean (r = 0.82, n = 150 engagements). The rating differed by role, with individual contributors rating it lower than engagement managers and portfolio directors.

All 10 written responses were negative. Eight said no person senior to the respondent knew the agents or had run them, and four described a question passed from a lead to another person that never returned.

“When I ask about how tool decide, my Lead also dont know, he ask someone in other city and no answer come back… Nobody from company sat with us even one day since account start.”Individual Contributor, BPO, Trade, settlement and reconciliation

“Nobody in my line management has used one of these tools for a day — that is the honest position, and it lands on me.”Engagement Manager, CXM, Client servicing and communications

“My management chain struggles to understand basic AI realities. I bounce between meetings with executives that either think AI is overpromised or vastly under-respected.”Engagement Manager, ITO, Client servicing and communications

2.1.4 Manager Ability to Run Agentic Outsourcing Engagements

Respondents rated whether their manager knows how to manage agentic AI outsourcing engagements. The mean rating was 4.8 and the median 5.0 (n = 800 respondents), and 241 of the 800 answers (30.13%) were distressed. The association with the engagement indicator mean (r = 0.84, n = 150 engagements) equaled the strongest of the five manager indicators. Higher ratings were associated with lower intent to leave (r = −0.67, n = 752 respondents).

A single written response addressed the indicator, an anecdote with the strongest negative score. It placed the difference in a manager who had done the job before and could therefore judge it.

“On the old account my manager had done the job himself for years, so he knew when I said a batch was bad. Here they dont know, so they just push.”Individual Contributor, BPO, Corporate and back-office functions

2.1.5 Manager Ability to Run Agentic Engagements for Global Financial Institutions

Respondents rated the statement “My manager knows how to manage agentic AI outsourcing engagements for global banks.” The mean rating was 4.8 and the median 5.0 (n = 791 respondents), and 242 of the 791 answers (30.59%) were distressed. The rating was associated with all four study measures, including cancellation (r = −0.64, n = 150 engagements) and the engagement indicator mean (r = 0.84, n = 150 engagements). Canceled engagements returned a mean of 3.2 (n = 296 respondents).

No respondent wrote about the manager’s capability with global financial institutions in particular. The evidence for the indicator is the rating alone.

Job satisfaction was among the survey questions most strongly associated with cancellation. Respondent satisfaction (r = −0.81) and team satisfaction (r = −0.81) were two of the five survey questions with the strongest association, of the 172 for which the association was computed (n = 150 engagements). Nearly half of the ratings of the respondent’s own job were distressed, 365 of 780 answers (46.79%).

The written evidence ties unhappiness to the daily tasks of the engagement. Respondents described the same data cleaned every cycle, decisions made alone, colleagues who left and issues no longer raised because nothing changed.

“Frankly the part that wore us down was waiting. My people would sit on a misrouted ticket queue for a week while the two sides worked out who was allowed to change the routing rules, and in that time they are doing the triage by hand, again.”Engagement Manager, CXM, Corporate and back-office functions

“Honestly I am just tired now. Two of us spend the first hour of every shift re-checking what the agent did overnight, as such we start the real queue already behind. My teammate stop taking her break.”Individual Contributor, CXM, Regulatory, audit, tax and legal

Satisfaction measured against five years earlier had a lower distressed share, in 214 of 804 answers (26.62%) for the respondent and 242 of 797 (30.36%) for the team.

Respondents also named what sustained them. The question on what the respondent loves most about the job received 786 written answers from 786 respondents on all 150 engagements. Skills learned, colleagues and pay were the subjects named most, including by respondents on canceled engagements.

3.1 Current Satisfaction of the Respondent and the Team

Respondent satisfaction and team satisfaction were rated almost alike. Respondent satisfaction had a mean of 4.2 (n = 780 respondents) and team satisfaction a mean of 4.3 (n = 788 respondents). The distressed share was 365 of 780 answers (46.79%) for the respondent and 357 of 788 (45.30%) for the team.

Each of the two ratings held all four of its associations under each of the 22 robustness checks. Their associations with cancellation, r = −0.81 for each (n = 150 engagements), exceeded that of every one of the 100 rated indicators. Respondents on canceled engagements gave the lowest ratings of any segment.

The association of respondent satisfaction with intent to leave, r = −0.91 (n = 732 respondents), is the association of two answers from the same respondent about the same job. The association of team satisfaction with intent to leave was lower, r = −0.76 (n = 742 respondents).

3.1.1 Respondent Job Satisfaction

Respondents rated how happy they are with their current job. The mean rating was 4.2 and the median 4.0 (n = 780 respondents), and 365 of the 780 answers (46.79%) were distressed. Respondents on canceled engagements gave a mean of 1.8 (n = 300 respondents).

Happier respondents were on engagements with less cancellation (r = −0.81, n = 150 engagements) and higher customer satisfaction among active engagements (r = 0.76, n = 94 engagements). Their ratings were also associated with a higher engagement indicator mean (r = 0.86, n = 150 engagements) and lower intent to leave (r = −0.91, n = 732 respondents).

The 250 written responses came from 219 respondents on 113 engagements. Among the 124 responses outside the question on what the respondent loves most about the job, 60 said the respondent or the team was tired, exhausted or worn down. All four written outcome tests reached significance. Engagements with a negative written response had a higher cancellation rate, and respondents whose written responses were more negative reported higher intent to leave.

“Tired, mostly… two months and we still could not get a name for who owns the vendor master file at their end. We asked for access to their procurement system four times, and received a spreadsheet extract instead.”Engagement Manager, CXM, Corporate and back-office functions

“Its been more tiring than hard if that makes sense, we spend a lot of the day waiting on access to things and then catching up at the end. Nobody on my team is angry about it, we just come out of the week feeling like we did two jobs”Individual Contributor, CXM, Corporate and back-office functions

“How this has felt for months now is exhausting — closing IT tickets bot could not resolve and rerouting HR cases by hand, because there is still no written escalation path.”Engagement Manager, CXM, Corporate and back-office functions

3.1.2 Satisfaction of the Extended Team

Respondents rated how happy their extended team is with the current engagement. The mean rating was 4.3 and the median 4.0 (n = 788 respondents), and 357 of the 788 answers (45.30%) were distressed.

Higher ratings were associated with less cancellation (r = −0.81, n = 150 engagements) and a higher engagement indicator mean (r = 0.87, n = 150 engagements). They were also associated with higher customer satisfaction among active engagements (r = 0.68, n = 94 engagements) and lower intent to leave (r = −0.76, n = 742 respondents).

The 56 written responses came from 54 respondents on 44 engagements, and 43 were negative. Among the 47 responses outside the question on what the respondent loves most, 34 said the team was tired, exhausted or worn down. Ten said team members had left, were applying elsewhere or had asked to move.

“If every income calculation has to be opened and checked anyway, the processors were doing their old job plus a review job, and they knew it. Two of my best people asked to be moved within the first month.”Engagement Manager, BPO, Credit, underwriting and suitability

“so two of my analysts spent most of their evenings that month re-running tests instead of the work they were hired for. Please understand it is that, and not the volume, which wears people down. I lost one of them to another account over exactly this.”Engagement Manager, KPO, Regulatory, audit, tax and legal

“The effect on my team is tiredness more than anything. Every week we take decisions that are not ours to take, because waiting for both sides to agree costs more time than the ticket has. Nobody above notices, they only see the ticket time going up, and then we get asked why the numbers slipped”Individual Contributor, CXM, Corporate and back-office functions

3.2 Satisfaction Relative to Five Years Earlier

Satisfaction measured against five years earlier had a lower distressed share than current satisfaction. The respondent’s comparison was distressed in 214 of 804 answers (26.62%) and the team’s in 242 of 797 (30.36%), against 46.79% and 45.30% for current satisfaction on bases of 780 and 788.

Both comparisons were associated with all four study measures, with a weaker association with cancellation than current satisfaction showed. The correlation was r = −0.59 for each comparison (n = 150 engagements). The comparison is consistent with dissatisfaction tied to the present engagement more than to the course of the respondents’ working lives.

3.2.1 Respondent Satisfaction Against Five Years Earlier

Respondents rated the statement “Compared to 5 years ago, I am happier with the work I am doing now.” The mean rating was 4.9 and the median 5.0 (n = 804 respondents), and 214 of the 804 answers (26.62%) were distressed. Higher ratings were associated with less cancellation (r = −0.59, n = 150 engagements) and higher customer satisfaction among active engagements (r = 0.72, n = 94 engagements). They were also associated with lower intent to leave (r = −0.76, n = 755 respondents).

A single written response addressed the comparison, an anecdote. It described an AI tool that took the repeated questions and left the respondent the cases that needed judgment.

“before this i am answering same password and leave-balance question fifty time a day. now only the unusual ones reach me, where policy is not clear. my team is less bored and less tired by evening.”Individual Contributor, CXM, Corporate and back-office functions

3.2.2 Team Satisfaction Against Five Years Earlier

Respondents rated the statement “Compared to 5 years ago, the team I work with seems to be happier.” The mean rating was 4.9 and the median 5.0 (n = 797 respondents), and 242 of the 797 answers (30.36%) were distressed. Higher ratings were associated with a higher engagement indicator mean (r = 0.80, n = 150 engagements) and lower intent to leave (r = −0.61, n = 751 respondents).

The single written response on the team’s comparison is the response quoted on the respondent’s own comparison, from the same respondent. It described a team less bored and less tired by evening once an AI tool took the repeated questions.

3.3 Challenges Named as Reducing Job Satisfaction

The survey asked each respondent to name the top three challenges that reduce their job satisfaction, and 755 respondents on all 150 engagements answered about their own engagement. The answers named the preparation of the vendor and the financial institution more than the agents themselves. The subjects named most were the vendor’s delivery playbook (139 answers), the financial institution’s legacy systems (128), the pricing of agentic work (115) and the provisioning of access and agent identities (115).

The vendor’s commercial model followed with 110 answers and the supply of qualified agentic talent with 86. Review capacity against agent output was named in 77 answers and the explainability of agent output in 72. Unclear liability or regulatory accountability for agent errors was named in 72 answers.

“No real data governance before automating … Vendor-risk model was not built for this … Agent decisions were a black box to us”Individual Contributor, BPO, Corporate and back-office functions

“No real agentic playbook to follow yet … Data governance still catching up … Documentation always trails the actual process”Individual Contributor, BPO, Corporate and back-office functions

3.4 Fatigue in the Written Answers

Fatigue and resignation ran through the written answers. Across all written answers about the respondent’s own engagement, the state “tired” was expressed in 133 answers from 111 respondents on 74 engagements. The state “resigned” was expressed in 155 answers from 124 respondents on 76 engagements, and “frustrated” in 119 answers from 99 respondents on 71 engagements.

“it wear you down. Every region of theirs book a break different way, so we learn one and then the next month it change again. My team is four people and two already ask to move.”Individual Contributor, BPO, Trade, settlement and reconciliation

Resignation took the form of issues no longer raised. Respondents described waiting for access, decisions and data that did not come, and expectations lowered until the missing item was absorbed as part of the job.

“In all honesty i stopped expecting anything. Three months and my access to their break system never come through, we work off screenshots somebody email us each morning, nobody ask about that part”Individual Contributor, BPO, Trade, settlement and reconciliation

“Wears you down explaining the same thing to their people every week”Individual Contributor, CXM, Client servicing and communications

“Truthfully I am tired of being middle man. My team asks me every week which expense methodology is approved and I have no answer to give them, because their change board has not met since March. It wears on people”Engagement Manager, BPO, Trade, settlement and reconciliation

“I dont raise it as a defect because nobody actually fixes the root problem, were just told to keep the queue moving, its bloody exhausting.”Individual Contributor, BPO, Corporate and back-office functions

The state “calm or fine” was expressed in 115 answers from 94 respondents on 65 engagements, and “proud” in 22 answers. The states “tired” and “resigned” each appeared in more answers than “calm or fine.”

3.5 Sources of Sustained Satisfaction

The survey asked each respondent what they love most about their job, and 786 respondents on all 150 engagements answered about their own engagement. Skills learned were named in 427 answers, colleagues in 393 and pay in 214. Answers that paired a positive with a qualification revealing a failure numbered 204.

Respondents on canceled engagements valued the same subjects. The 296 answers from respondents on the 56 canceled engagements named skills learned in 238, colleagues in 192 and pay in 134.

“Learned pricing and margining lessons that will stick … Some solid people on a doomed project … Pay held up through the cancellation”Individual Contributor, BPO, Trade, settlement and reconciliation

“Learned more about liability frameworks than I expected … Coworkers who understood the mess we were in … The role stays interesting even when the project fails”Individual Contributor, BPO, Corporate and back-office functions

The use case itself was valued where respondents found the problem worth solving. The 26 answers that named the use case described the problem as real and the domain as worth learning.

“the security discipline this work has taught me … a genuinely cautious, well-designed IAM use case … colleagues who take the identity risks as seriously as I do”Engagement Manager, ITO, Technology engineering and operations

Trust and care were associated with every study measure, whether the respondent was rating the financial institution, the manager or the employer. Each of the 12 trust and care indicators was associated with cancellation, customer satisfaction, the engagement indicator mean and intent to leave, and every association held under each of the 22 robustness checks. The correlations with the engagement indicator mean ranged from r = 0.76 to r = 0.84 (n = 150 engagements).

The frequency of contact with the financial institution was the exception. Regular contact with peers, with the financial institution’s engagement management and with its leadership was distressed in between 5.19% and 6.01% of answers, on bases of 333 to 809 respondents. None of the three was associated with any study measure, and each association was equivalent to zero.

The contact result is a finding. Respondents talked with the financial institution’s staff, and the difficulties they described lay in access, data, sign-off and ownership, matters that contact at peer level could not resolve.

“Their integration team knows the product properly, when i ask about a webhook retry they come back the same day and they know what they are talking about. The gap is only on data.”Individual Contributor, CXM, Technology engineering and operations

4.1 Trust in and Care From the Financial Institution

Four indicators describe the respondent’s relationship with the financial institution. They cover trust in it, its care for the respondent, and the respondent’s knowledge of how it regarded the team and the engagement. Each had a mean of 4.9 and a median of 5.0, on bases of 789 to 799 respondents. About a quarter of the answers to each were distressed.

Engagement-level correlations rest on 150 engagements, or the 94 active engagements for customer satisfaction, and the correlations with intent to leave on 742 to 751 respondents.

The four moved together. Their associations with intent to leave, between r = −0.76 and r = −0.78, were stronger than their associations with cancellation, between r = −0.59 and r = −0.61. The written evidence on the relationship came almost entirely from respondents who valued it.

4.1.1 Trust in the Financial Institution

Respondents rated the statement “I trust the customer.” The mean rating was 4.9 and the median 5.0 (n = 799 respondents), and 204 of the 799 answers (25.53%) were distressed. Higher ratings were associated with lower intent to leave (r = −0.78, n = 750 respondents) and higher customer satisfaction among active engagements (r = 0.74, n = 94 engagements). Respondents on canceled engagements gave a mean of 3.5 (n = 296 respondents).

The 54 written responses came from 54 respondents on 37 engagements, and 51 were positive. All but two answered the question on what the respondent loves most about the job and named the relationship with the financial institution among the three subjects listed. The three negative responses described a relationship that soured or trust that was low.

“Client relationship here is stable enough to have honest conversations”Portfolio Director, CXM, Technology engineering and operations

4.1.2 Financial Institution Care for the Respondent

Respondents rated the statement “The customer knows and cares about me.” The mean rating was 4.9 and the median 5.0 (n = 789 respondents), and 216 of the 789 answers (27.38%) were distressed. Higher ratings were associated with less cancellation (r = −0.61, n = 150 engagements) and a higher engagement indicator mean (r = 0.78, n = 150 engagements). The 301 respondents on canceled engagements gave the statement a mean of 3.5.

No respondent wrote about whether the financial institution knew or cared about the respondent.

4.1.3 Knowledge of How the Financial Institution Regards the Team

Respondents rated the statement “I know how the customer feels about my team.” The mean rating was 4.9 and the median 5.0 (n = 798 respondents), and 212 of the 798 answers (26.57%) were distressed. Higher ratings were associated with lower intent to leave (r = −0.77, n = 751 respondents) and less cancellation (r = −0.59, n = 150 engagements). Its mean among respondents on canceled engagements was 3.5 (n = 298 respondents).

Two written responses addressed the indicator, both positive and both anecdotes. Each said the financial institution trusted the team, expressed as room to design and redesign the agentic process.

“Client trusts the team enough to let us redesign as we go”Individual Contributor, BPO, Trade, settlement and reconciliation

4.1.4 Knowledge of How the Financial Institution Regards the Engagement

Respondents rated the statement “I know how the customer feels about the engagement I am on.” The mean rating was 4.9 and the median 5.0 (n = 791 respondents), and 219 of the 791 answers (27.69%) were distressed. Higher ratings were associated with higher customer satisfaction among active engagements (r = 0.74, n = 94 engagements) and a higher engagement indicator mean (r = 0.78, n = 150 engagements). On canceled engagements, 305 respondents gave a mean of 3.6.

No respondent wrote about how the financial institution regarded the engagement.

4.2 Contact With the Financial Institution

Contact with the financial institution was the one family of relationship indicators equivalent to zero against all four study measures. Regular contact with peers inside the financial institution was distressed in 48 of 809 answers (5.93%), with its engagement management in 41 of 790 (5.19%) and with its leadership in 20 of 333 (6.01%). Respondents on canceled engagements rated contact about the same as respondents on active engagements.

Frequent contact did not resolve the difficulties the respondents described. The written responses on contact with peers placed the difficulty in access, data, sign-off or ownership, matters that the financial institution’s staff at peer level could not decide. The negative written responses on contact came from respondents who had met no person on the financial institution’s side at any level.

4.2.1 Regular Contact With Financial Institution Peers

Respondents rated the statement “I talk with peers inside the customer regularly.” The mean rating was 5.5 and the median 6.0 (n = 809 respondents), and 48 of the 809 answers (5.93%) were distressed. The rating was not associated with cancellation (r = 0.00, n = 150 engagements) or customer satisfaction among active engagements (r = −0.06, n = 94 engagements). The engagement indicator mean (r = −0.08, n = 150 engagements) and intent to leave (r = 0.01, n = 760 respondents) showed no association either, and each association was equivalent to zero.

The 48 written responses came from 47 respondents on 37 engagements, most from the comment field on the financial institution’s organization. In 27 responses, respondents said the financial institution’s people answered quickly or worked alongside the team, and placed the difficulty in access, data, sign-off or ownership. The eight responses with the strongest negative score came from respondents who never met a person on the financial institution’s side.

“I never speak to anyone from the client side, not once in eight months. So these questions are asking me to rate people I only know from the file names on a deal folder.”Individual Contributor, KPO, Corporate and back-office functions

4.2.2 Regular Contact With Financial Institution Engagement Management

Respondents rated the statement “I talk with engagement management inside the customer regularly.” The mean rating was 5.5 and the median 6.0 (n = 790 respondents), and 41 of the 790 answers (5.19%) were distressed. The rating was not associated with cancellation (r = 0.01, n = 150 engagements) or intent to leave (r = 0.05, n = 743 respondents), and each of its four associations was equivalent to zero.

The 26 written responses came mostly from the comment fields on the financial institution’s organization and on joint management. Seven described a standing call or a shared list kept with a manager on the financial institution’s side. All nine negative responses received the strongest negative score and came from respondents who never met a person in the financial institution’s management.

The written outcome tests point the other way from the rating. Engagements with a negative written response were canceled at a higher rate, and the respondents who wrote more negatively reported higher intent to leave. These tests rest on nine responses and describe a complete absence of contact, not its frequency.

“Their side has no single owner for this programme, so we run a standing Thursday call with the three compliance leads and take whatever we can get agreed in that hour as the decision of record.”Engagement Manager, BPO, Financial crime and client onboarding

4.2.3 Regular Contact With Financial Institution Leadership

Respondents rated the statement “I talk with the customer’s leadership team regularly,” and 333 respondents answered. The mean rating was 5.6 and the median 6.0 (n = 333 respondents), and 20 of the 333 answers (6.01%) were distressed. Contact with leadership showed no association with cancellation (r = 0.01, n = 150 engagements) or intent to leave (r = 0.07, n = 324 respondents). All four of its associations were equivalent to zero.

The 12 written responses are thin evidence. The eight negative responses received the strongest negative score and came from respondents who never met a person on the financial institution’s side. Three described direct contact with a senior leader or committee, arising after a problem had already occurred.

“half these questions assume somebody on both sides is managing this together. there was no joint anything here, we had a standup nobody from the bank joined.”Individual Contributor, ITO, Payments, cards and disputes

4.3 Trust in and Care From the Manager

Care from the manager meant backing in the written responses. Three indicators cover the manager’s care for the respondent, the manager’s care for the larger team and the respondent’s trust in the manager. Their distressed shares ranged from 243 of 804 answers (30.22%) to 247 of 797 answers (30.99%). Their associations with the engagement indicator mean ranged from r = 0.82 to r = 0.84 (n = 150 engagements).

Written responses on care for the larger team numbered 73, from 71 respondents on 57 engagements, and 72 were positive. A manager who backed the team was described in 41 of the 66 responses to the question on what the respondent loves most about the job. Respondents described a manager who owned an agent’s error and did not pass it down, including on engagements that failed.

“To be fair, this platform work is new for everybody here… I would not say they know agents deeply yet, but when the agent posted a wrong instruction on an account my lead owned it and did not push it down to us. That counts.”Individual Contributor, BPO, Corporate and back-office functions

“A manager who is actually fair given how tense the vendor talks get”Individual Contributor, BPO, Corporate and back-office functions

4.3.1 Manager Care for the Respondent

Respondents rated the statement “My manager cares about me.” The mean rating was 4.7 and the median 5.0 (n = 804 respondents), and 243 of the 804 answers (30.22%) were distressed. Higher ratings were associated with less cancellation (r = −0.65, n = 150 engagements) and a higher engagement indicator mean (r = 0.82, n = 150 engagements). The mean was 3.2 among respondents on canceled engagements (n = 306 respondents).

Seven written responses addressed the indicator, all positive and all from the question on what the respondent loves most about the job. Five named a manager who had the respondent’s back.

“Manager genuinely has my back”Individual Contributor, CXM, Corporate and back-office functions

4.3.2 Manager Care for the Larger Team

Respondents rated the statement “My manager cares about the larger team.” The mean rating was 4.7 and the median 5.0 (n = 797 respondents), and 247 of the 797 answers (30.99%) were distressed. Higher ratings were associated with higher customer satisfaction among active engagements (r = 0.73, n = 94 engagements) and lower intent to leave (r = −0.68, n = 748 respondents). Respondents on canceled engagements gave a mean of 3.1 (n = 304 respondents).

The 73 written responses came from 71 respondents on 57 engagements, and 72 were positive. Every one of the 66 responses to the question on what the respondent loves most about the job named the manager. With a single negative response, no written outcome test reached significance, and the associations rest on the ratings.

“My employer has been decent with me personally, my lead backs us when a flag turns out to be wrong and that matters more than people think. We are short handed though, two people left and weren’t replaced, so the rota is tight.”Individual Contributor, CXM, Credit, underwriting and suitability

4.3.3 Trust in the Manager

Respondents rated the statement “I trust my manager.” The mean rating was 4.7 and the median 5.0 (n = 807 respondents), and 245 of the 807 answers (30.36%) were distressed. Its association with the engagement indicator mean, r = 0.84 (n = 150 engagements), was the strongest of the three manager care and trust indicators. Ratings on canceled engagements averaged 3.2 (n = 314 respondents).

Two written responses addressed the indicator, both positive, a base of anecdote strength. Each rested trust on the manager’s honesty, in one case on an engagement the respondent otherwise found had little to offer.

“very little on this engagement … the process knowledge I picked up … a manager who was at least honest”Individual Contributor, CXM, Corporate and back-office functions

4.4 Trust in and Care From the Employer

Five indicators describe the respondent’s relationship with the employer. They cover its care for the respondent and the team, its top leadership’s care for the respondent’s wellbeing and development, trust in it, and its effective use of the respondent. The distressed share of each lay between 192 of 786 answers (24.43%) and 218 of 799 (27.28%).

The bases are 150 engagements for cancellation and the engagement indicator mean, 94 active engagements for customer satisfaction, and 740 to 762 respondents for intent to leave.

The written responses on the employer describe care as attention, training and backing. Its absence was described as being left alone with the tasks of the engagement, with no manager senior to the lead who knew the respondent’s name.

“The problem is nobody above my lead knows my name… no career path was ever discussed about with me. I like the work, I like my team. But I have no idea what I am in two years.”Individual Contributor, ITO, Technology engineering and operations

4.4.1 Employer Care for the Respondent

Respondents rated the statement “My employer cares about me.” The mean rating was 5.0 and the median 5.0 (n = 786 respondents), and 192 of the 786 answers (24.43%) were distressed, the lowest share among the five employer indicators. Higher ratings were associated with lower intent to leave (r = −0.77, n = 740 respondents) and less cancellation (r = −0.56, n = 150 engagements).

Six written responses addressed the indicator, all from the comment field on the employer, a base of anecdote strength. The four negative responses described respondents left alone, with no training, no person who listened and no person who checked in.

“Two months of this and I stopped caring, which is not like me. Nobody checked in once”Individual Contributor, ITO, Regulatory, audit, tax and legal

4.4.2 Employer Care for the Team

Respondents rated the statement “My employer cares about the team.” The mean rating was 4.9 and the median 5.0 (n = 802 respondents), and 214 of the 802 answers (26.68%) were distressed. Its associations with customer satisfaction among active engagements (r = 0.65, n = 94 engagements) and intent to leave (r = −0.59, n = 751 respondents) were the weakest among the five employer indicators.

Four written responses addressed the indicator, split evenly between positive and negative, a base of anecdote strength. In each, care for the team turned on training and on backing when something went wrong.

“Am comfortable with how the firm has backed this team, training budget was approved without argument and two of my people sat the agent operations course in full.”Engagement Manager, BPO, Corporate and back-office functions

“we are on our own here”Individual Contributor, CXM, Corporate and back-office functions

4.4.3 Top Leadership Care for Wellbeing and Development

Respondents rated the statement “My employer’s top leadership team cares about my wellbeing and development.” The mean rating was 4.9 and the median 5.0 (n = 812 respondents), and 214 of the 812 answers (26.35%) were distressed. Respondents who reported that care also reported lower intent to leave (r = −0.76, n = 762 respondents), and their engagements had higher customer satisfaction among active engagements (r = 0.74, n = 94 engagements). On canceled engagements the rating had a mean of 3.6 (n = 303 respondents).

No respondent wrote about whether the employer’s top leadership cared about the respondent’s wellbeing and development.

4.4.4 Trust in the Employer

Respondents rated the statement “I trust my employer.” The mean rating was 4.9 and the median 5.0 (n = 799 respondents), and 218 of the 799 answers (27.28%) were distressed. Its association with customer satisfaction among active engagements, r = 0.79 (n = 94 engagements), was the strongest among the five employer indicators. Canceled engagements recorded a mean of 3.5 across 300 respondents.

No respondent wrote about trust in the employer.

4.4.5 Effective Use of the Respondent by the Employer

Respondents rated the statement “I am being used effectively by my employer.” The mean rating was 4.9 and the median 5.0 (n = 808 respondents), and 212 of the 808 answers (26.24%) were distressed. Higher ratings were associated with less cancellation (r = −0.61, n = 150 engagements) and lower intent to leave (r = −0.78, n = 756 respondents). Among the 307 respondents on canceled engagements, the mean rating was 3.5.

The 33 written responses came from 33 respondents on 31 engagements, and 21 were negative. Eight described a team too small for its tasks or teammates who left and were not replaced, and six said the respondent spent the day on tasks the respondent was not hired for. The 11 responses to the question on what the respondent loves most named a steady pace or workload.

“What the ratings do not show is how alone we are on this account. Our own platform team is good on paper but they support ten accounts, so when routing breaks on a Friday I am the escalation and the fix both.”Engagement Manager, CXM, Technology engineering and operations

“my whole afternoon is review now. two of us do nothing else some days, and it is not the work i was hired for”Individual Contributor, ITO, Client servicing and communications

“it say job completed and half the composites are not built. so nobody on my team trust that summary now, we open the output and count accounts by hand. it add two hours to my evening, that is not what i was hired for.”Individual Contributor, KPO, Trade, settlement and reconciliation

“Spent most of the month sitting in calls where neither side would decide anything. Three of us on this pod and two of us were doing nothing for days at a time. It wears you down more than being busy does”Individual Contributor, CXM, Technology engineering and operations

Intent to leave is one of the four core study measures. A rating of 7 or above on the 10-point scale of likelihood to leave the current job within 12 months was given by 430 of 798 respondents (53.88%). The distressed range, the answers nearest certain departure, held 370 of the 798 answers (46.37%).

Intent to leave was concentrated on canceled engagements. All 300 ratings from respondents on canceled engagements were distressed. Those respondents answered after their engagement had ended, and the result records a job already ending as much as an intention.

“Asking about two years from now is a joke, i do not know if i am here next month”Individual Contributor, KPO, Corporate and back-office functions

“I have four analysts on this and two of them have asked me this month to be moved elsewhere. We chase their compliance side for a threshold decision, nothing comes back, and my people sit idle and then work late when it finally arrives.”Engagement Manager, CXM, Financial crime and client onboarding

“the team was never staffed with anybody who has done this provisioning work before. i am the third person on this queue in three months. And that is the main thing.”Individual Contributor, ITO, Technology engineering and operations

The preference for working directly for a financial institution was associated with every poor outcome. The two statements on that preference have no good or bad end. Stronger agreement with each was associated with more cancellation, lower customer satisfaction, a lower engagement indicator mean and higher intent to leave.

The departures have a consequence the ratings do not measure. Written responses on team intent to leave described the knowledge of the manual corrections leaving with the people who made them, unwritten and unavailable to the staff who remained.

5.1 Intent to Leave of the Respondent and the Team

Respondent and team intent to leave are rated on the goodness view, the reversed scale on which a higher figure is the better outcome and a distressed rating marks a respondent likely to leave. Respondent intent was distressed in 370 of 798 answers (46.37%) and team intent in 368 of 799 (46.06%).

The active engagements held intent to leave at a lower level. In Build-Out, 19 of 166 ratings (11.45%) were distressed, in Transition/Cut-Over 39 of 237 (16.46%) and in Production 12 of 95 (12.63%). Respondents on canceled engagements gave 300 of the 370 distressed ratings in the study set.

Both measures were associated with the other study measures. On the goodness view, respondent intent to leave was associated with cancellation (r = −0.80, n = 150 engagements) and the engagement indicator mean (r = 0.84, n = 150 engagements). It was also associated with customer satisfaction among active engagements (r = 0.73, n = 94 engagements). Team intent to leave was associated with cancellation at r = −0.81 (n = 150 engagements).

5.1.1 Respondent Intent to Leave Within Twelve Months

Respondents rated how likely they are to leave their current job in the next 12 months. On the goodness view, the mean rating was 4.2 and the median 4.0 (n = 798 respondents), and 370 of the 798 answers (46.37%) were distressed. Every rating from a respondent on a canceled engagement was distressed (n = 300 respondents, mean 1.7 on the goodness view).

The indicator is itself the measure of intent to leave, and no association with intent to leave was computed. On the goodness view, higher ratings were associated with less cancellation (r = −0.80, n = 150 engagements) and a higher engagement indicator mean (r = 0.84, n = 150 engagements). Every association held under each of the 16 robustness checks that apply to the measure.

Ten written responses addressed the indicator. Four said the respondent was already looking for another job or did not expect to stay, and three named pay as what kept the respondent in the job.

“My lead say escalate but escalate to who, nothing come back. Two people on my team already resign and honestly i am also looking.”Individual Contributor, CXM, Corporate and back-office functions

“Pay is the main thing holding me here … Picked up real QA instincts … A few teammates worth keeping in touch with”Individual Contributor, CXM, Corporate and back-office functions

5.1.2 Team Intent to Leave Within Twelve Months

Respondents rated how likely members of their current team are to leave their jobs in the next 12 months. On the goodness view, the mean rating was 4.3 and the median 4.0 (n = 799 respondents), and 368 of the 799 answers (46.06%) were distressed. Respondents on canceled engagements gave a mean of 1.8 on the goodness view (n = 308 respondents).

On the goodness view, higher ratings were associated with less cancellation (r = −0.81, n = 150 engagements), the second strongest association with cancellation among the 172 survey questions. They were also associated with a higher engagement indicator mean (r = 0.86, n = 150 engagements) and lower respondent intent to leave (r = −0.76, n = 749 respondents).

The 55 written responses came from 52 respondents on 45 engagements. Every one of the 50 responses outside the question on what the respondent loves most described team members who had left, were looking for other work or had asked to move. In 13 of the 50, the knowledge of the manual corrections left with the people who made them.

“My analysts came onto this thinking they will learn something and instead they spend six months rewriting research notes by hand while nobody above decides anything. Two of them are already asking me for a move.”Engagement Manager, KPO, Corporate and back-office functions

“the effect on us is two extra hours most evenings checking what it produced, and one of our seniors left in may partly for this. i dont blame him.”Individual Contributor, ITO, Technology engineering and operations

“my teammate left in march for the same reason exactly. the ones who stay are the ones who stopped caring and i can feel myself going that way, which is the part i actually mind.”Individual Contributor, ITO, Technology engineering and operations

“Last Tuesday i sit two hours on one alert because nobody will say if i can close it myself. That is what wears you. i check dispositions in the evening on my phone now. Two of my team already ask to move off.”Individual Contributor, ITO, Financial crime and client onboarding

“Two of my best analysts resigned since March. I spend my days on status calls where both firms agree there is a problem and neither will own it, then I go back and ask the same people to give up another weekend.”Engagement Manager, BPO, Corporate and back-office functions

5.2 Job Seeking Relative to Five Years Earlier

Job seeking measured against five years earlier was distressed in fewer answers than current intent to leave. On the goodness view, the respondent’s comparison was distressed in 216 of 801 answers (26.97%) and the team’s in 221 of 799 (27.66%). Current intent was distressed in 370 of 798 answers (46.37%) for the respondent and 368 of 799 (46.06%) for the team.

Both comparisons were associated with all four study measures, with weaker associations with cancellation than current intent showed, r = −0.59 for the respondent and r = −0.57 for the team (n = 150 engagements). No respondent wrote about either comparison.

5.2.1 Respondent Job Seeking Against Five Years Earlier

Respondents rated the statement “Compared to 5 years ago, I am more likely to seek a new job.” On the goodness view, the mean rating was 4.9 and the median 5.0 (n = 801 respondents), and 216 of the 801 answers (26.97%) were distressed. Higher ratings on the goodness view were associated with higher customer satisfaction among active engagements (r = 0.74, n = 94 engagements) and lower intent to leave (r = −0.76, n = 754 respondents).

No respondent wrote about the comparison.

5.2.2 Team Job Seeking Against Five Years Earlier

Respondents rated the statement “Compared to 5 years ago, the team I work with is more likely to seek a new job.” On the goodness view, the mean rating was 4.9 and the median 5.0 (n = 799 respondents), and 221 of the 799 answers (27.66%) were distressed. Higher ratings on the goodness view were associated with a higher engagement indicator mean (r = 0.78, n = 150 engagements) and lower intent to leave (r = −0.60, n = 754 respondents).

No respondent wrote about the team’s comparison.

5.3 Commitment to the Financial Industry and Preference for Financial Institution Employment

Respondents remained committed to the financial industry and leaned toward working for the financial institution itself. The commitment to continue working on projects in the global financial industry was distressed in 214 of 784 answers (27.30%). The two statements of preference for direct employment by a financial institution had means of 6.0 (n = 799 respondents) and 6.1 (n = 802 respondents) on a scale with no good or bad end.

The preference was associated with every poor outcome. Stronger agreement with each statement was associated with more cancellation, lower customer satisfaction among active engagements, a lower engagement indicator mean and higher intent to leave. Respondents on canceled engagements gave means of 7.5 (n = 295 respondents) and 7.4 (n = 305 respondents).

No respondent wrote about either statement. One written response on team departures recorded a team that left the vendor to work for the financial institution, and the cost of that departure to the engagement.

“The team that did the work left to work for the bank. I had no understanding of how much they did beyond their defined job to keep things running. A few days after they left, the FX agent failed. I had to call the bank and ask them to send the team back to us for a few days to fix it.”Portfolio Director, BPO, Corporate and back-office functions

5.3.1 Commitment to Work in the Global Financial Industry

Respondents rated the statement “I want to continue working on projects in the global financial industry.” The mean rating was 4.9 and the median 5.0 (n = 784 respondents), and 214 of the 784 answers (27.30%) were distressed. Higher ratings were associated with less cancellation (r = −0.57, n = 150 engagements) and lower intent to leave (r = −0.75, n = 739 respondents).

No respondent wrote about the commitment.

5.3.2 Preference for Employment in a Financial Institution Capability Center

Respondents rated the statement “I would prefer to work directly for a bank in one of their global capability centers.” The statement has no good or bad end, and its direction is the finding. The mean rating was 6.0 and the median 6.0 (n = 799 respondents).

Stronger agreement was associated with more cancellation (r = 0.61, n = 150 engagements) and lower customer satisfaction among active engagements (r = −0.74, n = 94 engagements). It was also associated with a lower engagement indicator mean (r = −0.78, n = 150 engagements) and higher intent to leave (r = 0.79, n = 752 respondents). Respondents on canceled engagements gave a mean of 7.5 (n = 295 respondents), far higher than the means on the three active stages.

No respondent wrote about the preference.

5.3.3 Preference for Direct Financial Institution Employment Over Vendor Employment

Respondents rated the statement “Working directly for a bank would be better than working for an outsourcing vendor.” The rating scale for the statement has no good or bad end. The mean rating was 6.1 and the median 6.0 (n = 802 respondents).

Stronger agreement was associated with more cancellation (r = 0.59, n = 150 engagements) and higher intent to leave (r = 0.75, n = 753 respondents). Its association with the engagement indicator mean, r = −0.79 (n = 150 engagements), was the strongest of its four associations. Respondents on canceled engagements gave a mean of 7.4 (n = 305 respondents).

The comparison received no written response.

5.4 Sentiment Toward the Manual Alterations

Respondents who wrote about their manual alterations described them as routine. The six written-response questions on manual alteration received 2,325 answers about the respondent’s own engagement, from 641 respondents on 146 engagements. An alteration made every cycle or most days was described in 559 answers from 372 respondents on 134 engagements. A workaround that had become the standard way of performing the task was described in 143 answers from 129 respondents.

The emotional state expressed most in those answers was resignation, in 145 answers from 117 respondents on 76 engagements. Frustration appeared in 104 answers and tiredness in 87. A calm or neutral state appeared in 82 answers, and pride in 6.

“i just fix it myself and move on… dont discuss about it with anyone, no need to, we all do same thing quietly and i think that is the real process now, works better than the version somebody wrote down anyway.”Individual Contributor, CXM, Corporate and back-office functions

Some respondents framed the alteration as skill. Competence or technical achievement was described in 62 answers from 59 respondents.

“I don’t make a big thing of it, I just fix it and send it through, quicker that way, I’m good at catching what’s wrong before it moves on.”Individual Contributor, BPO, Corporate and back-office functions

Others had stopped raising the underlying problem. A problem raised that produced no response was described in 233 answers from 187 respondents, and raising it as futile in 134 answers from 111 respondents. The reasons, the methods and the disclosure of the alterations are measured in full with manual alteration (Part 5).

“last sprint upstream Api moved and we are last to know, again. i just patch data myself, dont even ask anymore, whats the point.”Individual Contributor, ITO, Technology engineering and operations

The manual practice that sustained these engagements was held in people. The written responses describe corrections made by habit, workarounds kept in private notes and fallback methods known to a few staff. The people who held that knowledge were among those most likely to leave.

“the manual runbook for a network outage needs rewriting and rehearsing, since the four people who used to absorb that by hand have gone, and what is left is a document nobody has opened in a year.”Engagement Manager, BPO, Trade, settlement and reconciliation

“On the agent side there are two people in the whole floor who really know the tooling, and both are shared across accounts.”Individual Contributor, CXM, Technology engineering and operations

Intent to leave was high among individual contributors, the role that performs the manual corrections. A rating of 7 or above on likelihood to leave was given by 342 of 616 individual contributors (55.52%). Respondents likely to leave had somewhat shorter careers than the others, a small difference bound to role.

The knowledge leaves with the people. Among the 50 negative written responses on team intent to leave, 7 described a manual workaround held by habit or a check kept in a person’s head. The financial institution’s protection when an agent or a provider failed rested on that knowledge (Part 2).

6.1 Experience Among Respondents Likely to Leave

Respondents likely to leave, those who rated their likelihood of leaving at 7 or above, were compared with respondents not likely to leave on years of career experience. The comparison rests on the 746 respondents who answered both questions.

6.1.1 Shorter Careers Among Respondents Likely to Leave

Respondents likely to leave reported fewer years of career experience than respondents not likely to leave. The mean was 3.77 years (n = 409 respondents) against 4.44 years (n = 337 respondents), a difference of −0.66 years. The difference reached significance by Welch’s test (t = −2.04, p = 0.041) and with standard errors clustered by engagement (t = −2.64, df = 149, p = 0.009, 150 engagements).

The effect is small (Cohen’s d = −0.15). Across the 746 respondents with both answers, years of experience correlated with intent to leave at r = −0.09 (p = 0.002, clustered by engagement). Years of experience track role closely, and the difference cannot be separated from role.

6.2 Undocumented Manual Practice Held by Departing Staff

The manual practice was undocumented in each form the written responses describe. Workarounds operated as habits and not as controls. Corrections went unlogged, documentation trailed what had been built, and the skill to perform a task by hand eroded as staff left (Part 2).

“The seniors who used to do the correspondent matching by hand have left, and nobody wrote it down. thus if a feed changes format on a Friday there is no one on my shift who knows the old method, we simply hold the alerts till Monday”Individual Contributor, BPO, Financial crime and client onboarding

“Two people used to fix its date formats by hand every morning and both have left, and nobody wrote down what they were doing.”Engagement Manager, ITO, Client servicing and communications

The knowledge passed between colleagues instead. Respondents described learning the agents and the financial domain from the person beside them, from a departing colleague and from a shared file of notes that new joiners read before the official material.

“So my team make our own note file, every time somebody find how to fix a case we put it there. Now new joiner read that file first, not the official one. That is how we learn here”Individual Contributor, BPO, Corporate and back-office functions

“I learn trust administration from the person sitting next to me, not from any training. And when she leaves, that knowledge goes also.”Individual Contributor, BPO, Corporate and back-office functions

“And the training was for the ticket tool, not for procurement, so I learned what a PO status actually means from the guy sitting next to me.”Individual Contributor, CXM, Corporate and back-office functions

A correction held only in a person’s habit stops when the person leaves. The delivery team’s manual alterations, measured in their reach, reasons and disclosure, are the same corrections at risk of departure (Part 5).

“What would help practically is a written list of the small corrections the analysts used to do on the bureau file before the model saw it, name matching mostly. Two of those people have moved on and it is only in their heads”Individual Contributor, KPO, Credit, underwriting and suitability